Solvation of Halide Ions by Methanol and 2,2,2-Trifluoroethanol
253
An Infrared Study of the Solvation of Halide Ions by Methanol and 2,2,2-Trifluoroethanol Ralph R. Ryall, Howard A. Strobel,' Paul M. Gross Chemical Laboratory, Duke University, Durham, North Carolina 27706
and Martyn C. R. Symons Department of Chemistry, The University, Leicester, LE1 7RH, England (Received June 3, 1976)
Using infrared spectroscopy, we have measured the concentration of free and hydrogen bonded methanol in carbon tetrachloride solutions of tetraalkylammonium halides (C1-,Br ,I-)and calculated formation constants and free energies for the halide monosolvates. Formation constants have also been obtained for trifluoroethanol and these anions. Further, for the iodide-methanol solvate, enthalpy and entropy values of formation have also been determined. For both alcohols, frequency shifts from the (O-H)free absorption on solvation are proportional to the free energy of hydrogen-bond formation, and to the half-widths of the bands. Fluoride ion was included in this part of the study and, as expected, the order of shifts was F . > C1- > Br- > I . Formation constants for trifluoroethanol solvates were an order of magnitude larger, reflecting the greater acidity of this alcohol. The spectral shifts on complex formation were also proportionately enhanced.
Introduction Although the 0-H stretching region of the infrared spectra for hydroxylic solvents is of limited use for studying ionic solvation, because the breadth of the features involved is usually greater than the band shifts, valuable information can nevertheless be obtained by studying dilute solutions in relatively inert solvents.'y2 Certain tetraalkylammonium salts dissolve in inert solvents such as carbon tetrachloride or methylene chloride as ion pairs or clusters, and added protic solvents selectively hydrogen bond to the anions. The 0-H stretching frequencies of monomeric alcohol molecules appear as narrow bands whose oscillator strengths can be readily determined, and these can be fairly accurately monitored to provide the total concentration of unbound alcohol. The anion-alcohol complexes give broader bands a t lower frequencies, the shift being a function of the hydrogen-bond strengths. There have been several previous studies of methanol-halide ion complexes, but the results are not very consistent, and we have therefore carried out a careful study of these complexes in carbon tetrachloride in order to obtain accurate results for comparison with our results for the 2,2,2-trifluoroethanol systems. In 1958, Lund3 established the pattern of spectral changes when alcohols solvate tetraalkylammonium halides in chloroform solution. Bufalini and Stern4 measured formation constants a t various temperatures for the bromide salt and methanol using benzene as an inert solvent, and hence calculated AH" and AS" parameters. They were particularly concerned with the presence of ion--quadrupoles and higher clusters, and concluded that ion pairs were solvated in preference to clusters. Hyne and Levy," on the other hand, using tert-butyl alcohol and the bromide salt in carbon tetrachloride discussed the spectral changes in terms of aggregation of the solvent rather than hydrogen bonding to the bromide ion. Allerhand and Schleyer' studied all four halide ions and concluded that the shifts (Av) from the free 0-H frequency followed the unexpected order C1 > F > Br- > I . They explained this result in terms of a uniquely strong interaction between the tetraalkylammonium cations and fluoride anions. Blandamer et studied the interaction between alkylammonium iodides and methanol and correlated the results +ith those from ultraviolet spectroscopic studies of the CTTS band for the iodide ions. Mohr et a1.2 ex-
tended these studies to water in carbon tetrachloride and found the expected trend in Av of F- > C1- > Br > I-..] Lipovskii and Dem'yanova7 using tetradecylammonium halides in carbon tetrachloride also found that fluoride ion induced the greatest shift. They explained the previous results' in terms of the presence of water. It is extremely difficult to remove water from alkylammonium fluorides without inducing decomposition. One very interesting aspect of these studies7 was the comparison between tetraalkylammonium and trialkylammonium salts. The Av and AGO values for the latter were considerably smaller than those for the former. Experimental Section Materials. Carbon Tetrachloride. Fisher ACS certified carbon tetrachloride was purified by fractional distillation, with only the middle fraction being collected. A Karl Fischer titration indicated not more than 0.01 % water after distillation. Methanol. Fisher ACS certified methanol was dried over magnesium. When there was no further evolution of hydrogen, the methanol was fractionally distilled and the middle portion collected. The purified methanol was found to contain no more than 0.03% water. 2,2,2-Trifluoromethanol. Eastman White Label 2,2,2-trifluoroethanol was used without further purification. Gas chromatographic and infrared analysis gave no indication of the presence of water or other impurities. T e t r a h e x y l a m m o n i u m Iodide and T e t r a b u t y l ammonium Bromide. Eastman Kodak salts were dried in vacuo at 75 "C for 4 days. A Karl Fischer titration on methanolic solutions indicated that not more than 0.02% water was present. Tetrabutylammonium Chloride. Tetrabutylammonium chloride was prepared from silver chloride and tetrabutylammonium iodide. The product was dried in vacuo at 7 5 "C. Analysis revealed the absence of significant amounts of iodide or water. Tetraalkylammonium Fluorides. We were unable to obtain dry samples of any tetraalkylammonium fluorides despite many attempts using a variety of procedures. However, we were able to obtain values for Y ( O H )for ~ MeOH---F adducts (ca. 3240 cm-') by dissolving the salts in dry methanol after removing as much water as possible, and drying the resulting solutions over molecular sieves The Journal of Physical Chemistry, Vol. SI, No, 3, 1977
254
R. R. Ryall, H. A. Strobel, and M. C. R. Symons
TABLE I: Molar Absorptivities, e(OH)free,for Monomeric Methanol and 2,2,2-Trifluoroethanol in CCI,, together with Approximate Values of €(OH),,for Methanol-Halide Adducts MeOH ( 1 0 "C)
d
/
4
e(OH)freer 53.5 i. 4 M-' cm-'
/
/
MeOH
MeOH (45 "C)
CF,CH,OH (29 "C)
43.9 * 4
33.4 f 3
103
MeOH- - -C1-
MeOH- - -Br-
MeOH- - -1-
185
174
122
4OH)J ( 2 9 C)
/
(29 "C)
decreased in intensity and a new band, which was much more broad and intense, appeared a t a lower frequency. The C-H stretch vibrations of the tetraalkylammonium cations gave rise to strong absorption bands, appearing around 3000 cm-l, which overlapped the broad 0-H stretching bands of the alcohol-halide complexes. Appropriate concentrations of the salt were added to the reference solution to compensate for these bands. The equilibrium between the alcohol and the tetraalkylammonium salt can be represented by the equation: I
I
I
I
O,w5
0,OM
0,015
O,@M
h 1 T Y OF k W L
(M3L L-')
Figure 1. Absorbance of v(OH), for methanol and 2,2,2-trifluoroethanol as a function of concentration in carbon tetrachloride at 29 "C.
for a prolonged period. These solutions were concentrated by vacuum distillation, and then added to carbon tetrachloride. The results were reproducible and the addition of traces of water resulted in the development of a new band at higher frequencies. Spectrometry. Infrared spectrophotometric measurements were made using sealed, matched 1.00-mm sodium chloride cells supplied by Wilks Scientific Corp. By measuring the interference fringe patterns obtained on running spectra of the empty cells, the cells were found to be matched to within 3%. These were housed in a thermostatted cell chamber for studies a t temperatures other than ambient (29 f 1 "C). All infrared measurements were performed on a Perkin-Elmer P-E 621 spectrometer. The ordinate spectral presentation was linear in percent transmittance. The spectral slit width was 3 cm-l which corresponds to ca. 117 of the half-band width of the monomeric methanol peak. Samples were prepared by weight, using a drybox when necessary. Peak heights of the u(OH) free band were used to estimate concentrations after it was established that Beer's law was obeyed (Figure 1). The concentration of free methanol was determined directly and that of bound methanol found by difference on addition of the halide. Ordinate scale expansion of 5 X was used as necessary. One major difficulty in measuring either the peak height or area was choosing the proper baseline. In this work the baseline was drawn as a straight line connecting the regions of constant transmission on either side of the absorption band. The true baseline was not strictly horizontal through the spectral region under investigation. As has been stressed by Potts,s uncertainties in the position of the baseline can result in larger errors for integrated absorbance measurements than for peak height measurements.
Results Dilute alcohol solutions in carbon tetrachloride gave a sharp, apparently Gaussian peak slightly above 3600 cm-l. When a quaternary ammonium salt was added, this peak The Journal of Physical Chemistrv, Vol. 81, No. 3, 1977
ROH
+ (R,N+)X-
I(ROH),
ROH- - - -X'(R,N+)
+ n(R4N'X-),
= [(RdN'X-),-
(1)
- - -(HOR),],
where R is an alkyl group and X is a halide. It is assumed that I = n = 1; i = lz = 1;and p = q = m = 1. Thus, the equilibrium constant, K , can be calculated from the following equation K = [ (R4N')X-- - -HOR]/[ ROH] [ (R4N')X-] For the dilute solutions studied, activity coefficients are assumed to be equal to unity. All concentration terms were corrected when necessary for changes in density with temperature. Finally, values for the Gibbs free energy change in formation of the solvate, AGO, were calculated using the mean value for the equilibrium constant at a specific temperature. The enthalpy of formation of the hydrogen bond between Hex,NI and methanol was estimated from the temperature dependence of the equilibrium constant as given by the van't Hoff equation:
An enthalpy value, AH", of -3.7 kcal/mol was determined by plotting R In K against 1/T. Further, using the relation, AGO = AH"- T A S " , the entropy of hydrogen bond formation was found to be -6.7 eu for the complex a t 29 "C. The concentration plots establishing adherence to Beer's law are shown in Figure 1. The resulting molar absorpat 29 "C are given in Table I, together with tivities e(OH)free those for methanol a t 10 and 45 "C. Also included are approximate values for e(OH)b, the molar absorptivities for the bound 0-H group. The values of e(OH), were calculated from the equilibrium concentration of the alcohol-halide adduct and the peak absorbance for the hydrogen bonded complex. For systems at 29 "C the calculated formation constants and AGO values are listed in Table I1 together with band frequencies v(OH), and band shifts. The frequencies of band maxima were in agreement with those reported by Allerhand and Schleyerl and Lipov~kii.~Calculated thermodynamic data for methanol iodide adducts are given in Table 111. Formation constants are the mean of approximately ten determinations covering a range of total salt concentration between ca. 0.02 and 0.1 M for methanol and ca. 6.001 and 0.05 M for 2,2,2-trifluoroethanol. The total concentration
255
Solvation of Halide Ions by Methanol and 2,2,2-Trifluoroethanol TABLE 11: Formation Constants, A G o Values, and Spectral Parameters for Alcohol-Halide Ion Complexes in Carbon Tetrachloride at 29 + 1 C Avu(OH)f,w AGO, 40H)f,,e, v(OHh,, (OHh? '~(OH-jii, Alcohol Salt K kcal/mol cm-l cm-' cm-l cm-' MeOH
CF,CH,OH~
Bu,NF Bu,NCl Bu,NBr Hex,NI
1 7 3 + 16 49 i. 6 16.8 f 3
Bu,NCl Bu,NBr Hex,NI
7660 r 500 920 + 160 1 7 0 + 10
-2.3 -1.7
3643 3643 3643 3643
(3240) 3275 3339 3390
185 145 120
(403) 368 304 253
-5.4 -4.1 - 3.1
3621 3621 3621
3140 3220 3285
230 207 180
481 401 336
-- 3.1
Data applies to those cases where the ratio of initial concentration of Bu,NBr or Band width at half maximum. Hex,NI to the concentration of alcohol was greater than 1 . 5 ; for Bu,NCI the ratio was greater than unity. TABLE 111: Thermodynamic Data for Methanol-Iodide Solvates Temp, "C ~
A G O ,
AH",
K
kcal/mol
kcal/mol
A S " , eu
27.1 i 2.4 16.8 i. 2.5 13.2 i 1.2
-1.86 -1.69 -1.63
-3.7
-6.7
~~
10 29 45
of methanol or 2,2,2-trifluoroethanol was generally ca. 0.02 M. The magnitudes of the shifts from the u(OH)freu maxima are shown as a function of the free energies, AGO, in Figure 2. The values for u(OH), were independent of concentration in the range covered except for the fluoroalcohol adducts where shifts to higher frequencies were detected when the [alcohol] > [salt]. These shifts are discussed below. Also u(OH),,, for methanol and U(OH)b for MeOH---I - adducts were independent of temperature between 10 and 45 OC. The K values showed no trends with concentration within the concentration ranges used.
Discussion Our values for the molar absorptivity for methanol, agree reasonably with previous value^.^^"' However, it is clearly not temperature independent, as implied by Lipovskii and Dem'yan~va.~ Also, our values for t(OH)b, given in Table I, do not agree well with those reported for their CgHI90H---hal adduct^.^ Our AH value of -3.7 kcal/mol for MeOH---I- is smaller than that of -5.3 reported for C9HI90H---I.7 This may be because of the difference in the alcohols, but since the latter was apparently based on the assumption of a temperature independent value for t(OH)free, it is probably slightly large. On this basis, the difference in A S (-6.7 eu for MeOH---I- and -10.5 for CgHI90H---I-')is also understandable. The formation constant for MeOH---I reported by Blandamer et al.6 of 13.7 is in fair agreement with the present value, but the Bufalini and Stern value for MeOH---Br in benzene4 of 65.2 at 25 "C is higher than our value (49.0 at 29 "C). However, in this study, Bufalini and Stern used dielectric data to calculate equilibrium concentrations, and made allowance for the presence of quadrupoles on the assumption that only the ion pairs interacted with methanol. We see no reason why the alcohol molecules should discriminate between ion pairs and ion-quadrupoles, and this is supported by the selfconsistency of the present results. Our infrared results for fluoride, presented in Table 11, confirm the findings of Lipovskii and Dem'yanova7 that u(OH)b for MeOH---F- is located at a lower frequency than the band maximum for MeOH---Cl-. This order would be expected from the relative charge densities of the halide ions. It thus appears that the anomalous infrared results
' J ( O H ) ~- ~V~( O~ H ) ~ (cm-')
Figure 2. Free energy of formation of various hydrogen-bonded complexes at 29 O C vs. the shift, u(OH)+, - u(0HA: OH-I-, methanol (O), F3CCH20H(8);OH-Br-, methanol (A),FBCCHpOH(A):OH-CI-, methanol (O),F3CCH20H (0).
obtained by Allerhand and Schleyer' for MeOH---F- may well be attributed to small amounts of water in their samples. They prepared the fluoride salt in an aqueous system. Because of the deliquescent nature of tetraalkylammonium fluorides we obtained only an estimate of K(Me0H---F ), from which it appears that K(MeOH---F ) > K(MeOH---Cl.). Results for the fluoroalcohol accord well with its increased acidity and indeed give a measure of this increase. It is of some interest that the results in Figure 2 for the two alcohols are nearly colinear. We do not think that the deviations from a single curve have any special significance. Though not shown, a similar, also essentially linear plot is obtained by graphing the half-widths Au(OH)b of the solvate bands for the two alcohols against the frequency shifts for the solvates. The shifts to high frequencies in v(OH)b as the [F$CH,OH] is increased above that for the salt is interpreted in terms of the addition of a second molecule: ROH- - -Hal-
+ ROH
